Author: Stuzzi Kadent
Date: 01:26:12 09/13/02
Go up one level in this thread
On September 12, 2002 at 17:08:55, John Merlino wrote: >On September 12, 2002 at 16:25:06, Terry Ripple wrote: > >>The company tested 5 personalities in CM 8000 with established rated players. >>They played 3,000 human games against the 5 personalities and 10,500 additional >>games(100 test games for each personality) were played using computer vrs. >>computer matches with the other personalities with in Chessmaster to fill out >>the ratings of all the other personalities. This was all done for CM 8000, but >>now that we have a totally new product called CM 9000 it would be interesting to >>get players who have ratings to play against some of those personalities >>once again and if time warrants we could even test more than just five, but that >>depends on how many human players(rated) would be interested to play some >>personalities as mentioned. >> >>They used a Fischer time control of 5/10 to do the tests! >> >>Here are the 5 personalities tested with their established USCF ratings: >>CM8000(2500), Josh(age9)2000, Willie(1500), Sonja(1000 )Skippy(5000). >> >>Would anyone be interested to test and/or have suggestions? >> >>Regards, >> Terry > >Just to clarify, those were the approximate rating ranges that we wanted to test >with. After the testing was completed, several of those personalities had their >ratings significantly lowered. The ACTUAL displayed ratings of those >personalities in CM9000 may be very different, depending on hardware. On my >PIII-600, the above personalities are rated as follows in CM9000: > >Chessmaster -- 2767 >Josh Age 9 -- 1830 >Willow (not Willie) -- 1104 >Sonja -- 824 >Skippy -- 643 > >So, if we wanted to try the same test with appropriate CM9000 personalities, I >would suggest: > >Chessmaster, Rand (2003), Dylan (1547), Joey (1104) and Manny (500). > >It may be silly to test the lowest two, since I expect everybody here will be >able to win every game against them. The reason that we tested such low-rated >personalities with CM8000 was that there were several dozen children who were >used in the test. > >Specifically, though, if you are going to participate in this test, here are the >parameters: > >1. Fischer 5/10 time control >2. Play an even number of games against each opponent with both White and Black. >3. Play these games in the Tournament Room using the Rated Game feature. Do not >use the Game Room to play these games. >4. DO NOT USE PAUSE during the game. > >Many thanks, > >jm Is all this refining of calibration against weakened settings really necessary. Do people go around saying "I am a xxxx elo player" because Chessmaster tells them so, and should we be concerned by it? It is more important that users find an appropriate level to play against, and the Chessmaster has many levels to choose from. It is really only necessary to try them briefly. Now rating the Chessmaster against Garry Kasparov would be interesting.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.